Netdata Community

apps_groups.conf and exporter when process is not running

Environment

My OS:

Distributor ID: Debian
Description: Debian GNU/Linux 11 (bullseye)
Release: 11
Codename: bullseye

Problem/Question

I have process/application configured and reporting properly in my apps_group.conf and when that process is running, it’s properly using the exporter to send to my influxdb so I can graph in Grafana.

When I stop the process though, it doesn’t report anything anymore.

What I expected to happen

I’d expect to send 0 as it’s not running anymore or even a null or something. Right now it sends nothing about that app until it starts back up.

I’ll see “1” when the process is up, but when it down, no new data comes into Influx:

1636221466000000000 1
1636221476000000000 1
1636221486000000000 1
1636221496000000000 1
>
>
> SELECT "value" FROM "netdata.apps.processes.someprocessname" WHERE time >= now() - 5m and time <= now()
>

Is there something I can do so it can send 0s or whatnot to make sure I do get data for the times when the process is not running?

No, you can’t. It is intentional and hardcoded: if(unlikely(w->exposed && w->processes)). If there are no processes in a group, sending resource usage stats for this group would be misleading.

Could you please explain what is your use case? Why do you need that data?

Netdata->to a backend (opentsdb in this case)->Grafana. for visualization.

When the process stops, it can’t alert nor does it show up on the graph because there is no data at all that comes over so it makes any process data useless.

I did see a github issue stating something similar on how it works so I moved off netdata and just used monit->telegraf as that sends a status code of does not exist so graphing and all things work properly.

Appreciate the reply as I wanted to make sure I wasn’t missing anything.

We can send zeroes for apps.processes. It looks rational. @Animosity022, @ilyam8, what do you think?

That makes perfect sense to me as I’m not sure how edge my use case is but I feel it’s a pretty normal one as the combination of netdata->backend->some visualization tool seems common.